Discrepancy estimates based on Haar functions

From MaRDI portal





The author presents a technique to estimate the star-discrepancy of \((t,m,s)\)-nets using generalized Haar function systems, and applies it to prove that the star-discrepancy of \({\mathcal P}_t\), the digital \((t,m,2)\)-nets in base \(b=2\), satisfies \[ D_N^*({\mathcal P}_t)\leq 1- \biggl(1- \frac{1}{2^{m-t}} \biggr)^2+ \frac 13 \frac{m-2t}{2^m}+ \frac 19 \frac{1}{2^m}- \frac{(-1)^m}{9} \frac{1}{2^{2(m-t)}} \] for \(0\leq t\leq\lfloor m/2\rfloor\).











This page was built for publication: Discrepancy estimates based on Haar functions

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5938363)